From aa5a72fb3748aa1ecd8cf6e488661dd4b3108ed0 Mon Sep 17 00:00:00 2001
From: Oleksii Kliukin <oleksii.kliukin@zalando.de>
Date: Thu, 9 Jul 2015 15:45:10 +0200
Subject: [PATCH 01/32] Add on_start, on_stop, on_role_change, on_restart and
 on_reload callbacks.

---
 helpers/postgresql.py | 33 ++++++++++++++++++++++++++++++---
 1 file changed, 30 insertions(+), 3 deletions(-)

diff --git a/helpers/postgresql.py b/helpers/postgresql.py
index 560d7a98..a7c2f742 100644
--- a/helpers/postgresql.py
+++ b/helpers/postgresql.py
@@ -1,6 +1,7 @@
 import logging
 import os
 import psycopg2
+import shlex
 import shutil
 import subprocess
 import sys
@@ -42,6 +43,7 @@ class Postgresql:
         self.replication = config['replication']
         self.superuser = config['superuser']
         self.admin = config['admin']
+        self.callback = config.get(['callback'], {})
         self.recovery_conf = os.path.join(self.data_dir, 'recovery.conf')
         self.configuration_to_save = (os.path.join(self.data_dir, 'pg_hba.conf'),
                                       os.path.join(self.data_dir, 'postgresql.conf'))
@@ -238,6 +240,16 @@ class Postgresql:
     def is_running(self):
         return subprocess.call(' '.join(self._pg_ctl) + ' status > /dev/null', shell=True) == 0
 
+    def call_nowait(self, cb_name):
+        """ pick a callback command and call it without waiting for it to finish """
+        if not self.callback or cb_name not in self.callback:
+            return False
+        cmd = self.callback[cb_name]
+        is_master = self.is_leader()
+        name = self.name
+        subprocess.Popen(shlex.split(cmd)+[is_master, name])
+        return True
+
     def start(self):
         if self.is_running():
             self.load_replication_slots()
@@ -251,18 +263,29 @@ class Postgresql:
         ret = subprocess.call(self._pg_ctl + ['start', '-o', self.server_options()]) == 0
         ret and self.load_replication_slots()
         self.save_configuration_files()
+        if ret and 'on_start' in self.callback:
+            self.call_nowait('on_start')
         if self.on_change_callback:
             self.on_change_callback('replica' if os.path.exists(self.recovery_conf) else 'master')
         return ret
 
     def stop(self):
-        return subprocess.call(self._pg_ctl + ['stop', '-m', 'fast']) != 0
+        ret = subprocess.call(self._pg_ctl + ['stop', '-m', 'fast'])
+        if ret == 0 and 'on_stop' in self.callback:
+            self.call_nowait('on_stop')
+        return ret != 0
 
     def reload(self):
-        return subprocess.call(self._pg_ctl + ['reload']) == 0
+        ret = subprocess.call(self._pg_ctl + ['reload'])
+        if ret == 0 and 'on_reload' in self.callback:
+            self.call_nowait('on_reload')
+        return ret == 0
 
     def restart(self):
-        return subprocess.call(self._pg_ctl + ['restart', '-m', 'fast']) == 0
+        ret = subprocess.call(self._pg_ctl + ['restart', '-m', 'fast'])
+        if ret == 0 and 'on_restart' in self.callback:
+            self.call_nowait('on_restart')
+        return ret == 0
 
     def server_options(self):
         options = "--listen_addresses='{}' --port={}".format(self.listen_addresses, self.port)
@@ -349,6 +372,8 @@ primary_conninfo = '{}'
         if not self.check_recovery_conf(leader):
             self.write_recovery_conf(leader)
             self.restart()
+            if self.on_change_callback['on_role_change']:
+                self.call_nowait('on_role_change')
             if self.on_change_callback:
                 self.on_change_callback('replica')
 
@@ -370,6 +395,8 @@ primary_conninfo = '{}'
 
     def promote(self):
         self.is_promoted = subprocess.call(self._pg_ctl + ['promote']) == 0
+        if self.is_promoted and self.on_change_callback['on_role_change']:
+            self.call_nowait('on_role_change')
         if self.on_change_callback:
             self.on_change_callback('master')
         return self.is_promoted

From 2c1627c2f3aec4befc535c0239097ffccaf89671 Mon Sep 17 00:00:00 2001
From: Oleksii Kliukin <oleksii.kliukin@zalando.de>
Date: Fri, 10 Jul 2015 10:33:07 +0200
Subject: [PATCH 02/32] get rid of yaml module, since AWS instance-identity
 natively sends JSON.

---
 helpers/aws.py | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/helpers/aws.py b/helpers/aws.py
index b47e4e0f..26c65e89 100644
--- a/helpers/aws.py
+++ b/helpers/aws.py
@@ -1,8 +1,6 @@
 import logging
-import re
 import requests
 from requests.exceptions import RequestException
-import yaml
 import boto.ec2
 
 logger = logging.getLogger(__name__)
@@ -27,7 +25,7 @@ class AWSConnection:
             return
         if r.ok:
             try:
-                content = yaml.load(r.content)
+                content = r.json()
                 self.instance_id = content['instanceId']
                 self.region = content['region']
             except Exception as e:

From 78f0b16543541ce20b42c0ab97878b8b57b47d17 Mon Sep 17 00:00:00 2001
From: Oleksii Kliukin <oleksii.kliukin@zalando.de>
Date: Fri, 10 Jul 2015 10:57:24 +0200
Subject: [PATCH 03/32] Enable aws.py to run as a stand-alone program.

---
 helpers/aws.py | 13 +++++++++++++
 1 file changed, 13 insertions(+)

diff --git a/helpers/aws.py b/helpers/aws.py
index 26c65e89..02ddd7fe 100644
--- a/helpers/aws.py
+++ b/helpers/aws.py
@@ -1,6 +1,7 @@
 import logging
 import requests
 from requests.exceptions import RequestException
+import sys
 import boto.ec2
 
 logger = logging.getLogger(__name__)
@@ -67,3 +68,15 @@ class AWSConnection:
     def on_role_change(self, new_role):
         ret = self._tag_ec2(new_role)
         return self._tag_ebs(new_role) and ret
+
+
+if __name__ == '__main__':
+    if len(sys.argv) != 3:
+        print ("Usage: {0} action role name".format(sys.argv[0]))
+        return 1
+    action, role, name = sys.argv[1:]
+    if action in ('on_start', 'on_stop', 'on_role_change'):
+        aws = gAWSConnection({'cluster_name': name})
+        aws.on_role_change(role)
+        return 0
+    return 2

From c0cc59093820bd134935e508281e67e76647df6b Mon Sep 17 00:00:00 2001
From: Oleksii Kliukin <oleksii.kliukin@zalando.de>
Date: Fri, 10 Jul 2015 12:58:37 +0200
Subject: [PATCH 04/32] Remove usage of AWS.py from Patroni

Functionality provided by that module will be achieved via callbacks.
- fix a typo in aws.py
- fix some mixups of old code and new callbacks in postgresql.py
---
 helpers/aws.py        |  2 +-
 helpers/postgresql.py | 13 +++----------
 patroni.py            |  2 --
 3 files changed, 4 insertions(+), 13 deletions(-)

diff --git a/helpers/aws.py b/helpers/aws.py
index 02ddd7fe..6a04da92 100644
--- a/helpers/aws.py
+++ b/helpers/aws.py
@@ -76,7 +76,7 @@ if __name__ == '__main__':
         return 1
     action, role, name = sys.argv[1:]
     if action in ('on_start', 'on_stop', 'on_role_change'):
-        aws = gAWSConnection({'cluster_name': name})
+        aws = AWSConnection({'cluster_name': name})
         aws.on_role_change(role)
         return 0
     return 2
diff --git a/helpers/postgresql.py b/helpers/postgresql.py
index a7c2f742..5dff6578 100644
--- a/helpers/postgresql.py
+++ b/helpers/postgresql.py
@@ -35,7 +35,7 @@ def parseurl(url):
 
 class Postgresql:
 
-    def __init__(self, config, on_change_callback=None):
+    def __init__(self, config):
         self.config = config
         self.name = config['name']
         self.listen_addresses, self.port = config['listen'].split(':')
@@ -66,7 +66,6 @@ class Postgresql:
         self._connection = None
         self._cursor_holder = None
         self.members = []  # list of already existing replication slots
-        self.on_change_callback = on_change_callback
 
     def get_local_address(self):
         listen_addresses = self.listen_addresses.split(',')
@@ -265,8 +264,6 @@ class Postgresql:
         self.save_configuration_files()
         if ret and 'on_start' in self.callback:
             self.call_nowait('on_start')
-        if self.on_change_callback:
-            self.on_change_callback('replica' if os.path.exists(self.recovery_conf) else 'master')
         return ret
 
     def stop(self):
@@ -372,10 +369,8 @@ primary_conninfo = '{}'
         if not self.check_recovery_conf(leader):
             self.write_recovery_conf(leader)
             self.restart()
-            if self.on_change_callback['on_role_change']:
+            if 'on_role_change' in self.callback:
                 self.call_nowait('on_role_change')
-            if self.on_change_callback:
-                self.on_change_callback('replica')
 
     def save_configuration_files(self):
         """
@@ -395,10 +390,8 @@ primary_conninfo = '{}'
 
     def promote(self):
         self.is_promoted = subprocess.call(self._pg_ctl + ['promote']) == 0
-        if self.is_promoted and self.on_change_callback['on_role_change']:
+        if self.is_promoted and 'on_role_change' in self.callback:
             self.call_nowait('on_role_change')
-        if self.on_change_callback:
-            self.on_change_callback('master')
         return self.is_promoted
 
     def demote(self, leader):
diff --git a/patroni.py b/patroni.py
index d90e5c64..6f06a056 100755
--- a/patroni.py
+++ b/patroni.py
@@ -6,7 +6,6 @@ import time
 import yaml
 
 from helpers.api import RestApiServer
-from helpers.aws import AWSConnection
 from helpers.etcd import Etcd
 from helpers.ha import Ha
 from helpers.postgresql import Postgresql
@@ -18,7 +17,6 @@ class Patroni:
 
     def __init__(self, config):
         self.nap_time = config['loop_wait']
-        self.aws = AWSConnection(config)
         self.postgresql = Postgresql(config['postgresql'])
         self.ha = Ha(self.postgresql, self.get_dcs(self.postgresql.name, config))
         host, port = config['restapi']['listen'].split(':')

From 6d0a9bbdc0030734f415e61801bc40928735590f Mon Sep 17 00:00:00 2001
From: Oleksii Kliukin <oleksii.kliukin@zalando.de>
Date: Mon, 13 Jul 2015 10:25:42 +0200
Subject: [PATCH 05/32] create a new directory for callback scripts and move
 aws volume tags there.

---
 {helpers => scripts}/aws.py | 0
 1 file changed, 0 insertions(+), 0 deletions(-)
 rename {helpers => scripts}/aws.py (100%)

diff --git a/helpers/aws.py b/scripts/aws.py
similarity index 100%
rename from helpers/aws.py
rename to scripts/aws.py

From f734272f3f437bf0c38a92ff0183e4b2340bdd5a Mon Sep 17 00:00:00 2001
From: Oleksii Kliukin <oleksii.kliukin@zalando.de>
Date: Mon, 13 Jul 2015 11:45:25 +0200
Subject: [PATCH 06/32] Process the command paths relative to the patroni dir
 correctly, fix the error of not supplying a callback action to the actual
 command.

---
 helpers/postgresql.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/helpers/postgresql.py b/helpers/postgresql.py
index 5dff6578..de73ee93 100644
--- a/helpers/postgresql.py
+++ b/helpers/postgresql.py
@@ -246,7 +246,7 @@ class Postgresql:
         cmd = self.callback[cb_name]
         is_master = self.is_leader()
         name = self.name
-        subprocess.Popen(shlex.split(cmd)+[is_master, name])
+        subprocess.Popen(shlex.split(os.path.abspath(cmd))+[cb_name, is_master, name])
         return True
 
     def start(self):

From 15f94f42241d46b54cca38376fd1f927c28506e8 Mon Sep 17 00:00:00 2001
From: Oleksii Kliukin <oleksii.kliukin@zalando.de>
Date: Mon, 13 Jul 2015 11:47:55 +0200
Subject: [PATCH 07/32] Use actual role name and not boolean flag if it is a
 master to tag AWS objects (EBS volumes and EC2 instances).

---
 scripts/aws.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/scripts/aws.py b/scripts/aws.py
index 6a04da92..7930323e 100644
--- a/scripts/aws.py
+++ b/scripts/aws.py
@@ -74,9 +74,9 @@ if __name__ == '__main__':
     if len(sys.argv) != 3:
         print ("Usage: {0} action role name".format(sys.argv[0]))
         return 1
-    action, role, name = sys.argv[1:]
+    action, is_master, name = sys.argv[1:]
     if action in ('on_start', 'on_stop', 'on_role_change'):
         aws = AWSConnection({'cluster_name': name})
-        aws.on_role_change(role)
+        aws.on_role_change('master' if is_master else 'replica')
         return 0
     return 2

From d8aa1ff849989979e8bcfae3ae3b0e1f58237513 Mon Sep 17 00:00:00 2001
From: Oleksii Kliukin <oleksii.kliukin@zalando.de>
Date: Tue, 14 Jul 2015 09:22:49 +0200
Subject: [PATCH 08/32] fix a typo.

---
 helpers/postgresql.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/helpers/postgresql.py b/helpers/postgresql.py
index de73ee93..41bc13c8 100644
--- a/helpers/postgresql.py
+++ b/helpers/postgresql.py
@@ -43,7 +43,7 @@ class Postgresql:
         self.replication = config['replication']
         self.superuser = config['superuser']
         self.admin = config['admin']
-        self.callback = config.get(['callback'], {})
+        self.callback = config.get('callback', {})
         self.recovery_conf = os.path.join(self.data_dir, 'recovery.conf')
         self.configuration_to_save = (os.path.join(self.data_dir, 'pg_hba.conf'),
                                       os.path.join(self.data_dir, 'postgresql.conf'))

From fcead3aed94bc19e7ab2b5246484ae2670849783 Mon Sep 17 00:00:00 2001
From: Alexander Kukushkin <cyberdemn@gmail.com>
Date: Wed, 15 Jul 2015 15:25:20 +0200
Subject: [PATCH 09/32] Update README.md

---
 README.md | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/README.md b/README.md
index 3ecbcebb..b894c5d9 100644
--- a/README.md
+++ b/README.md
@@ -55,8 +55,11 @@ For an example file, see `postgres0.yml`.  Below is an explanation of settings:
   * *scope*: the relative path used on etcd's http api for this deployment, thus you can run multiple HA deployments from a single etcd
   * *session_timeout*: the TTL to acquire the leader lock. Think of it as the length of time before automatic failover process is initiated.
   * *reconnects_timeout*: how long we should try to reconnect to ZooKeeper after connection loss. After this timeout we assume that we don't have lock anymore and will restart in read-only mode.
-  * *hosts*: List of ZooKeeper cluster members in format: 'host1:port1,host2:port2,..etc...'
-
+  * *hosts*: list of ZooKeeper cluster members in format: [ 'host1:port1', 'host2:port2', 'etc...']
+  * *exhibitor*: if you are running ZooKeeper cluster under Exhibitor supervisory the following section could be interesting for you
+    * *poll_interval*: how often list of ZooKeeper and Exhibitor nodes should be updated from Exhibitor
+    * *port*: Exhibitor port
+    * *hosts*: initial list of Exhibitor (ZooKeeper) nodes in format: [ 'host1', 'host2', 'etc...' ]. This list would be updated automatically when Exhibitor (ZooKeeper) cluster topology changes.
 
 * *postgresql*
   * *name*: the name of the Postgres host, must be unique for the cluster

From d1e47c8ccfbcb298718be3e645dfb2b3644e9aad Mon Sep 17 00:00:00 2001
From: Alexander Kukushkin <cyberdemn@gmail.com>
Date: Wed, 15 Jul 2015 16:17:38 +0200
Subject: [PATCH 10/32] Update README.md

---
 README.md | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/README.md b/README.md
index b894c5d9..f90478ff 100644
--- a/README.md
+++ b/README.md
@@ -67,10 +67,17 @@ For an example file, see `postgres0.yml`.  Below is an explanation of settings:
   * *connect_address*: ip address + port through which Postgres is accessible from other nodes and applications.
   * *data_dir*: file path to initialize and store Postgres data files
   * *maximum_lag_on_failover*: the maximum bytes a follower may lag before it is not eligible become leader
+  * *pg_hba*: list of lines which should be added to pg_hba.conf
+    * *- host all all 0.0.0.0/0 md5*
   * *replication*
     * *username*: replication username, user will be created during initialization
     * *password*: replication password, user will be created during initialization
     * *network*: network setting for replication in pg_hba.conf
+  * *superuser*
+    * *password*: password for postgres user. It would be set during initialization
+  * *admin*:
+    * *username*: admin username, user will be created during initialization. It would have CREATEDB and CREATEROLE privileges
+    * *password*: admin password, user will be created during initialization.
   * *recovery_conf*: configuration settings written to recovery.conf when configuring follower
   * *parameters*: list of configuration settings for Postgres
 

From 11004f8c48062ebd3dea0bccda8985975e0217b7 Mon Sep 17 00:00:00 2001
From: Oleksii Kliukin <oleksii.kliukin@zalando.de>
Date: Fri, 17 Jul 2015 11:19:05 +0200
Subject: [PATCH 11/32] Be more robust on callbacks when cluster is down.

For 'on_stop' or 'on_restart' callbacks we cannot
call is_leader, because the cluster would be stopped.

Instead, call is_leader before running a callback.
In addition, make the call_nowait more robust by handling
the exceptions related to the DB cluster unavailability.
---
 helpers/postgresql.py | 25 ++++++++++++++++++++-----
 1 file changed, 20 insertions(+), 5 deletions(-)

diff --git a/helpers/postgresql.py b/helpers/postgresql.py
index 41bc13c8..4d72fcec 100644
--- a/helpers/postgresql.py
+++ b/helpers/postgresql.py
@@ -239,14 +239,19 @@ class Postgresql:
     def is_running(self):
         return subprocess.call(' '.join(self._pg_ctl) + ' status > /dev/null', shell=True) == 0
 
-    def call_nowait(self, cb_name):
+    def call_nowait(self, cb_name, is_leader=None):
         """ pick a callback command and call it without waiting for it to finish """
         if not self.callback or cb_name not in self.callback:
             return False
         cmd = self.callback[cb_name]
-        is_master = self.is_leader()
+        if not is_leader:
+            try:
+                is_leader = self.is_leader()
+            except psycopg2.OperationalError as e:
+                logger.warning("unable to perform {0} action, cannot obtain the cluster role: {1}".format(cb_name, e))
+                return False
         name = self.name
-        subprocess.Popen(shlex.split(os.path.abspath(cmd))+[cb_name, is_master, name])
+        subprocess.Popen(shlex.split(os.path.abspath(cmd))+[cb_name, is_leader, name])
         return True
 
     def start(self):
@@ -267,9 +272,14 @@ class Postgresql:
         return ret
 
     def stop(self):
+        try:
+            is_leader = self.is_leader()
+        except:
+            is_leader = None
+            pass
         ret = subprocess.call(self._pg_ctl + ['stop', '-m', 'fast'])
         if ret == 0 and 'on_stop' in self.callback:
-            self.call_nowait('on_stop')
+            self.call_nowait('on_stop', is_leader=is_leader)
         return ret != 0
 
     def reload(self):
@@ -279,9 +289,14 @@ class Postgresql:
         return ret == 0
 
     def restart(self):
+        try:
+            is_leader = self.is_leader()
+        except:
+            is_leader = None
+            pass
         ret = subprocess.call(self._pg_ctl + ['restart', '-m', 'fast'])
         if ret == 0 and 'on_restart' in self.callback:
-            self.call_nowait('on_restart')
+            self.call_nowait('on_restart', is_leader=is_leader)
         return ret == 0
 
     def server_options(self):

From 33b5e1b22e95a66f55819defc443ab4fa14f502f Mon Sep 17 00:00:00 2001
From: Oleksii Kliukin <oleksii.kliukin@zalando.de>
Date: Fri, 17 Jul 2015 15:22:21 +0200
Subject: [PATCH 12/32] Make sure is_leader is False does not trigger the
 leader re-check.

---
 helpers/postgresql.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/helpers/postgresql.py b/helpers/postgresql.py
index 4d72fcec..3fcb44da 100644
--- a/helpers/postgresql.py
+++ b/helpers/postgresql.py
@@ -244,7 +244,7 @@ class Postgresql:
         if not self.callback or cb_name not in self.callback:
             return False
         cmd = self.callback[cb_name]
-        if not is_leader:
+        if is_leader is None:
             try:
                 is_leader = self.is_leader()
             except psycopg2.OperationalError as e:

From a11d94aff53d00595e0e8c6c598a5af84a48d0e5 Mon Sep 17 00:00:00 2001
From: Oleksii Kliukin <oleksii.kliukin@zalando.de>
Date: Fri, 17 Jul 2015 15:39:34 +0200
Subject: [PATCH 13/32] Pass text names of cluster roles to the script.

Seems better than passing a bool flag and
converting it from/to string.
---
 helpers/postgresql.py | 2 +-
 scripts/aws.py        | 4 ++--
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/helpers/postgresql.py b/helpers/postgresql.py
index 3fcb44da..371c187a 100644
--- a/helpers/postgresql.py
+++ b/helpers/postgresql.py
@@ -251,7 +251,7 @@ class Postgresql:
                 logger.warning("unable to perform {0} action, cannot obtain the cluster role: {1}".format(cb_name, e))
                 return False
         name = self.name
-        subprocess.Popen(shlex.split(os.path.abspath(cmd))+[cb_name, is_leader, name])
+        subprocess.Popen(shlex.split(os.path.abspath(cmd))+[cb_name, "master" if is_leader else "replica", name])
         return True
 
     def start(self):
diff --git a/scripts/aws.py b/scripts/aws.py
index 7930323e..6a04da92 100644
--- a/scripts/aws.py
+++ b/scripts/aws.py
@@ -74,9 +74,9 @@ if __name__ == '__main__':
     if len(sys.argv) != 3:
         print ("Usage: {0} action role name".format(sys.argv[0]))
         return 1
-    action, is_master, name = sys.argv[1:]
+    action, role, name = sys.argv[1:]
     if action in ('on_start', 'on_stop', 'on_role_change'):
         aws = AWSConnection({'cluster_name': name})
-        aws.on_role_change('master' if is_master else 'replica')
+        aws.on_role_change(role)
         return 0
     return 2

From e9934be9aa80f04990eb07eb6708c50422cff4aa Mon Sep 17 00:00:00 2001
From: Oleksii Kliukin <oleksii.kliukin@zalando.de>
Date: Fri, 17 Jul 2015 17:26:28 +0200
Subject: [PATCH 14/32] Do not terminate on callback failure.

Show the error if a callback command fails and move on.
Set the executable bit for the aws callback.
---
 helpers/postgresql.py | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/helpers/postgresql.py b/helpers/postgresql.py
index 371c187a..0c31ddac 100644
--- a/helpers/postgresql.py
+++ b/helpers/postgresql.py
@@ -251,7 +251,12 @@ class Postgresql:
                 logger.warning("unable to perform {0} action, cannot obtain the cluster role: {1}".format(cb_name, e))
                 return False
         name = self.name
-        subprocess.Popen(shlex.split(os.path.abspath(cmd))+[cb_name, "master" if is_leader else "replica", name])
+        try:
+            role = "master" if is_leader else "replica"
+            subprocess.Popen(shlex.split(os.path.abspath(cmd))+[cb_name, role, name])
+        except Exception as e:
+            logger.warning("callback {0} {1} {2} {3} failed: {4}".format(os.path.abspath(cmd), cb_name, role, name, e))
+            return False
         return True
 
     def start(self):

From f1d76611b6b7c2f1b1a15c72976e5c1029f3b4a8 Mon Sep 17 00:00:00 2001
From: Oleksii Kliukin <oleksii.kliukin@zalando.de>
Date: Fri, 17 Jul 2015 17:26:52 +0200
Subject: [PATCH 15/32] Use the executable bit on AWS callback.

---
 scripts/aws.py | 0
 1 file changed, 0 insertions(+), 0 deletions(-)
 mode change 100644 => 100755 scripts/aws.py

diff --git a/scripts/aws.py b/scripts/aws.py
old mode 100644
new mode 100755

From c9c678fb5b8cbd80b806302d5d63159d1c31f922 Mon Sep 17 00:00:00 2001
From: Oleksii Kliukin <oleksii.kliukin@zalando.de>
Date: Thu, 23 Jul 2015 16:16:22 +0200
Subject: [PATCH 16/32] Pass scope, the cluster name, instead of the name,
 which is the instance name, to the callback.

---
 helpers/postgresql.py |  3 ++-
 scripts/aws.py        | 10 ++++++----
 2 files changed, 8 insertions(+), 5 deletions(-)
 mode change 100755 => 100644 scripts/aws.py

diff --git a/helpers/postgresql.py b/helpers/postgresql.py
index 0c31ddac..6348c3ff 100644
--- a/helpers/postgresql.py
+++ b/helpers/postgresql.py
@@ -38,6 +38,7 @@ class Postgresql:
     def __init__(self, config):
         self.config = config
         self.name = config['name']
+        self.scope = config['scope']
         self.listen_addresses, self.port = config['listen'].split(':')
         self.data_dir = config['data_dir']
         self.replication = config['replication']
@@ -250,7 +251,7 @@ class Postgresql:
             except psycopg2.OperationalError as e:
                 logger.warning("unable to perform {0} action, cannot obtain the cluster role: {1}".format(cb_name, e))
                 return False
-        name = self.name
+        name = self.scope
         try:
             role = "master" if is_leader else "replica"
             subprocess.Popen(shlex.split(os.path.abspath(cmd))+[cb_name, role, name])
diff --git a/scripts/aws.py b/scripts/aws.py
old mode 100755
new mode 100644
index 6a04da92..0025e44e
--- a/scripts/aws.py
+++ b/scripts/aws.py
@@ -1,3 +1,5 @@
+#!/usr/bin/python
+
 import logging
 import requests
 from requests.exceptions import RequestException
@@ -71,12 +73,12 @@ class AWSConnection:
 
 
 if __name__ == '__main__':
-    if len(sys.argv) != 3:
+    if len(sys.argv) != 4:
         print ("Usage: {0} action role name".format(sys.argv[0]))
-        return 1
+        sys.exit(1)
     action, role, name = sys.argv[1:]
     if action in ('on_start', 'on_stop', 'on_role_change'):
         aws = AWSConnection({'cluster_name': name})
         aws.on_role_change(role)
-        return 0
-    return 2
+        sys.exit(0)
+    sys.exit(2)

From 35fe536e8b6ff2a073682592243744c7d1b36f4a Mon Sep 17 00:00:00 2001
From: Oleksii Kliukin <oleksii.kliukin@zalando.de>
Date: Thu, 23 Jul 2015 17:09:39 +0200
Subject: [PATCH 17/32] Set +x bit on aws.py

---
 scripts/aws.py | 0
 1 file changed, 0 insertions(+), 0 deletions(-)
 mode change 100644 => 100755 scripts/aws.py

diff --git a/scripts/aws.py b/scripts/aws.py
old mode 100644
new mode 100755

From dac3d7997fdb2eb9994a0dfe2de543cb9de1c17b Mon Sep 17 00:00:00 2001
From: Oleksii Kliukin <oleksii.kliukin@zalando.de>
Date: Fri, 24 Jul 2015 10:47:26 +0200
Subject: [PATCH 18/32] Change a typo.

---
 helpers/postgresql.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/helpers/postgresql.py b/helpers/postgresql.py
index 6348c3ff..a5ae8b36 100644
--- a/helpers/postgresql.py
+++ b/helpers/postgresql.py
@@ -44,7 +44,7 @@ class Postgresql:
         self.replication = config['replication']
         self.superuser = config['superuser']
         self.admin = config['admin']
-        self.callback = config.get('callback', {})
+        self.callback = config.get('callbacks', {})
         self.recovery_conf = os.path.join(self.data_dir, 'recovery.conf')
         self.configuration_to_save = (os.path.join(self.data_dir, 'pg_hba.conf'),
                                       os.path.join(self.data_dir, 'postgresql.conf'))

From 859da0db0adb98505ad840afd0cd4f7527399554 Mon Sep 17 00:00:00 2001
From: Oleksii Kliukin <oleksii.kliukin@zalando.de>
Date: Mon, 27 Jul 2015 10:10:29 +0200
Subject: [PATCH 19/32] Make unit tets run again.

---
 postgres0.yml            |  6 ++++--
 postgres1.yml            |  6 ++++--
 tests/test_aws.py        | 32 +++++++++++++++++++++++++-------
 tests/test_postgresql.py | 10 +++++-----
 4 files changed, 38 insertions(+), 16 deletions(-)

diff --git a/postgres0.yml b/postgres0.yml
index 0cb694a1..6c9fb82d 100644
--- a/postgres0.yml
+++ b/postgres0.yml
@@ -1,20 +1,22 @@
 ttl: &ttl 30
 loop_wait: &loop_wait 10
+scope: &scope batman
 restapi:
   listen: 127.0.0.1:8008
   connect_address: 127.0.0.1:8008
 etcd:
-  scope: batman
+  scope: *scope
   ttl: *ttl
   host: 127.0.0.1:4001
   #discovery_srv: my-etcd.domain
 #zookeeper:
-#  scope: batman
+#  scope: *scope
 #  session_timeout: *ttl
 #  reconnect_timeout: *loop_wait
 #  hosts: 127.0.0.1:2181
 postgresql:
   name: postgresql0
+  scope: *scope
   listen: 127.0.0.1:5432
   connect_address: 127.0.0.1:5432
   data_dir: data/postgresql0
diff --git a/postgres1.yml b/postgres1.yml
index 680091f1..1928cee7 100644
--- a/postgres1.yml
+++ b/postgres1.yml
@@ -1,20 +1,22 @@
 ttl: &ttl 30
 loop_wait: &loop_wait 10
+scope: &scope batman
 restapi:
   listen: 127.0.0.1:8009
   connect_address: 127.0.0.1:8009
 etcd:
-  scope: batman
+  scope: *scope
   ttl: *ttl
   host: 127.0.0.1:4001
   #discovery_srv: my-etcd.domain
 #zookeeper:
-#  scope: batman
+#  scope: *scope
 #  session_timeout: *ttl
 #  reconnect_timeout: *loop_wait
 #  hosts: 127.0.0.1:2181
 postgresql:
   name: postgresql1
+  scope: *scope
   listen: 127.0.0.1:5433
   connect_address: 127.0.0.1:5433
   data_dir: data/postgresql1
diff --git a/tests/test_aws.py b/tests/test_aws.py
index 63b50734..4331badb 100644
--- a/tests/test_aws.py
+++ b/tests/test_aws.py
@@ -2,7 +2,7 @@ import unittest
 import requests
 import boto.ec2
 from collections import namedtuple
-from helpers.aws import AWSConnection
+from scripts.aws import AWSConnection
 from requests.exceptions import RequestException
 import yaml
 
@@ -24,6 +24,16 @@ class MockEc2Connection:
         return True
 
 
+class MockResponse:
+
+    def __init__(self, content):
+        self.content = content
+        self.ok = True
+
+    def json(self):
+        return self.content
+
+
 class TestAWSConnection(unittest.TestCase):
 
     def __init__(self, method_name='runTest'):
@@ -44,25 +54,28 @@ class TestAWSConnection(unittest.TestCase):
         result = namedtuple('Request', 'ok content')
         result.ok = True
         if url.split('/')[-1] == 'document':
-            result.content = '{\n  "instanceId" : "012345",\n "region" : "eu-west-1"\n}'
+            result = {"instanceId": "012345", "region": "eu-west-1"}
         else:
-            result.content = 'foo'
-        return result
+            result = 'foo'
+        return MockResponse(result)
 
     def setUp(self):
         self.error = False
         requests.get = self.requests_get
         boto.ec2.connect_to_region = self.boto_ec2_connect_to_region
         self.config_string = """
-loop_wait: 10
+scope: &scope test
+ttl: &ttl 30
+loop_wait: &loop_wait 10
 restapi:
   listen: 0.0.0.0:8008
   connect_address: 127.0.0.1:5432
 etcd:
-  scope: test
-  ttl: 30
+  scope: *scope
+  ttl: *ttl
   host: 127.0.0.1:8080
 postgresql:
+  scope: *scope
   name: postgresql_foo
   listen: 0.0.0.0:5432
   connect_address: 127.0.0.1:5432
@@ -76,6 +89,11 @@ postgresql:
   admin:
     username: admin
     password: admin
+  callbacks:
+    on_start: patroni/scripts/aws.py
+    on_stop: patroni/scripts/aws.py
+    on_restart: patroni/scripts/aws.py
+    on_role_change: patroni/scripts/aws.py
   parameters:
     archive_mode: "on"
     wal_level: hot_standby
diff --git a/tests/test_postgresql.py b/tests/test_postgresql.py
index d6c01f5b..069c4622 100644
--- a/tests/test_postgresql.py
+++ b/tests/test_postgresql.py
@@ -110,15 +110,15 @@ class TestPostgresql(unittest.TestCase):
     def set_up(self):
         subprocess.call = subprocess_call
         shutil.copy = nop
-        self.p = Postgresql({'name': 'test0', 'data_dir': 'data/test0', 'listen': '127.0.0.1, *:5432',
-                             'connect_address': '127.0.0.2:5432',
+        self.p = Postgresql({'name': 'test0', 'scope': 'batman', 'data_dir': 'data/test0',
+                             'listen': '127.0.0.1, *:5432', 'connect_address': '127.0.0.2:5432',
                              'pg_hba': ['hostssl all all 0.0.0.0/0 md5', 'host all all 0.0.0.0/0 md5'],
-                             'superuser': {'password': ''}, 'admin': {'username': 'admin', 'password': 'admin'},
+                             'superuser': {'password': ''},
+                             'admin': {'username': 'admin', 'password': 'admin'},
                              'replication': {'username': 'replicator',
                                              'password': 'rep-pass',
                                              'network': '127.0.0.1/32'},
-                             'parameters': {'foo': 'bar'}, 'recovery_conf': {'foo': 'bar'}},
-                             on_change_callback=lambda state: True)
+                             'parameters': {'foo': 'bar'}, 'recovery_conf': {'foo': 'bar'}})
         psycopg2.connect = psycopg2_connect
         if not os.path.exists(self.p.data_dir):
             os.makedirs(self.p.data_dir)

From 201f9bf4b29e3bd6cb9c9ccb4a25e62663d04b7a Mon Sep 17 00:00:00 2001
From: Oleksii Kliukin <oleksii.kliukin@zalando.de>
Date: Mon, 27 Jul 2015 11:08:19 +0200
Subject: [PATCH 20/32] Make sure role check during callbacks does not
 accidentially modify the cluster state (is_promoted flag or trigger file).

---
 helpers/postgresql.py | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/helpers/postgresql.py b/helpers/postgresql.py
index a5ae8b36..71eb7d9c 100644
--- a/helpers/postgresql.py
+++ b/helpers/postgresql.py
@@ -230,9 +230,9 @@ class Postgresql:
         return (diff_in_bytes < long(threshold_megabytes) * 1048576) and\
                (diff_in_bytes < long(backup_size) * float(threshold_backup_size_percentage) / 100)
 
-    def is_leader(self):
+    def is_leader(self, check_only=False):
         ret = not self.query('SELECT pg_is_in_recovery()').fetchone()[0]
-        if ret and self.is_promoted:
+        if ret and self.is_promoted and not check_only:
             self.delete_trigger_file()
             self.is_promoted = False
         return ret
@@ -247,7 +247,7 @@ class Postgresql:
         cmd = self.callback[cb_name]
         if is_leader is None:
             try:
-                is_leader = self.is_leader()
+                is_leader = self.is_leader(check_only=True)
             except psycopg2.OperationalError as e:
                 logger.warning("unable to perform {0} action, cannot obtain the cluster role: {1}".format(cb_name, e))
                 return False
@@ -279,7 +279,7 @@ class Postgresql:
 
     def stop(self):
         try:
-            is_leader = self.is_leader()
+            is_leader = self.is_leader(check_only=True)
         except:
             is_leader = None
             pass
@@ -296,7 +296,7 @@ class Postgresql:
 
     def restart(self):
         try:
-            is_leader = self.is_leader()
+            is_leader = self.is_leader(check_only=True)
         except:
             is_leader = None
             pass

From 115ba1d9ae2cf4c87526d5fcb2ccc92b58414fde Mon Sep 17 00:00:00 2001
From: Oleksii Kliukin <oleksii.kliukin@zalando.de>
Date: Mon, 27 Jul 2015 11:23:08 +0200
Subject: [PATCH 21/32] Make sure postgresql.stop also return True if pg_ctl
 stop returned 0.

---
 helpers/postgresql.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/helpers/postgresql.py b/helpers/postgresql.py
index 71eb7d9c..da2d1fdf 100644
--- a/helpers/postgresql.py
+++ b/helpers/postgresql.py
@@ -286,7 +286,7 @@ class Postgresql:
         ret = subprocess.call(self._pg_ctl + ['stop', '-m', 'fast'])
         if ret == 0 and 'on_stop' in self.callback:
             self.call_nowait('on_stop', is_leader=is_leader)
-        return ret != 0
+        return ret == 0
 
     def reload(self):
         ret = subprocess.call(self._pg_ctl + ['reload'])

From 56983d867db449c218af97766043b8d9966fb079 Mon Sep 17 00:00:00 2001
From: Oleksii Kliukin <oleksii.kliukin@zalando.de>
Date: Mon, 27 Jul 2015 13:03:12 +0200
Subject: [PATCH 22/32] Run coverage test for the data in scripts as well as
 helpers.

---
 setup.py | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/setup.py b/setup.py
index 5ac4a89f..db327c37 100644
--- a/setup.py
+++ b/setup.py
@@ -22,6 +22,7 @@ __location__ = os.path.join(os.getcwd(), os.path.dirname(inspect.getfile(inspect
 NAME = 'patroni'
 MAIN_PACKAGE = 'patroni.py'
 HELPERS = 'helpers'
+SCRIPTS = 'scripts'
 VERSION = '0.1'
 DESCRIPTION = 'A Template for PostgreSQL HA with etcd'
 LICENSE = 'The MIT License'
@@ -61,7 +62,8 @@ class PyTest(TestCommand):
     def finalize_options(self):
         TestCommand.finalize_options(self)
         if self.cov_xml or self.cov_html:
-            self.cov = ['--cov', MAIN_PACKAGE, '--cov', HELPERS, '--cov-report', 'term-missing']
+            self.cov = ['--cov', MAIN_PACKAGE, '--cov', HELPERS, '--cov', SCRIPTS, '--cov-report',
+                        'term-missing']
             if self.cov_xml:
                 self.cov.extend(['--cov-report', 'xml'])
             if self.cov_html:

From b92bfb34a59555523ca66172e908da697b3c961f Mon Sep 17 00:00:00 2001
From: Oleksii Kliukin <oleksii.kliukin@zalando.de>
Date: Mon, 27 Jul 2015 13:04:01 +0200
Subject: [PATCH 23/32] Futher test for postgresql module (includes callbacks)

---
 tests/test_postgresql.py | 20 ++++++++++++++++++--
 1 file changed, 18 insertions(+), 2 deletions(-)

diff --git a/tests/test_postgresql.py b/tests/test_postgresql.py
index 069c4622..c93a7a3c 100644
--- a/tests/test_postgresql.py
+++ b/tests/test_postgresql.py
@@ -118,7 +118,11 @@ class TestPostgresql(unittest.TestCase):
                              'replication': {'username': 'replicator',
                                              'password': 'rep-pass',
                                              'network': '127.0.0.1/32'},
-                             'parameters': {'foo': 'bar'}, 'recovery_conf': {'foo': 'bar'}})
+                             'parameters': {'foo': 'bar'}, 'recovery_conf': {'foo': 'bar'},
+                             'callbacks': {'on_start': '/usr/bin/true', 'on_stop': '/usr/bin/true',
+                                           'on_restart': '/usr/bin/true', 'on_role_change': '/bin/true',
+                                           'on_reload': '/usr/bin/true'
+                                           }})
         psycopg2.connect = psycopg2_connect
         if not os.path.exists(self.p.data_dir):
             os.makedirs(self.p.data_dir)
@@ -129,6 +133,9 @@ class TestPostgresql(unittest.TestCase):
     def tear_down(self):
         shutil.rmtree('data')
 
+    def mock_query(self, p):
+        raise psycopg2.OperationalError("not supported")
+
     def test_data_directory_empty(self):
         self.assertTrue(self.p.data_directory_empty())
 
@@ -136,12 +143,13 @@ class TestPostgresql(unittest.TestCase):
         self.assertTrue(self.p.initialize())
         self.assertTrue(os.path.exists(os.path.join(self.p.data_dir, 'pg_hba.conf')))
 
-    def test_start(self):
+    def test_start_stop(self):
         self.assertFalse(self.p.start())
         self.p.is_running = is_running
         with open(os.path.join(self.p.data_dir, 'postmaster.pid'), 'w'):
             pass
         self.assertTrue(self.p.start())
+        self.assertTrue(self.p.stop())
 
     def test_sync_from_leader(self):
         self.assertTrue(self.p.sync_from_leader(self.leader))
@@ -196,3 +204,11 @@ class TestPostgresql(unittest.TestCase):
 
     def test_last_operation(self):
         self.assertEquals(self.p.last_operation(), '0')
+
+    def test_non_existing_callback(self):
+        self.assertFalse(self.p.call_nowait('foobar'))
+
+    def test_is_leader_exception(self):
+        self.p.start()
+        self.p.query = self.mock_query
+        self.assertTrue(self.p.stop())

From d85451efae29d4d7a4e10ff36aa557d2f22556b5 Mon Sep 17 00:00:00 2001
From: Oleksii Kliukin <oleksii.kliukin@zalando.de>
Date: Mon, 27 Jul 2015 13:04:45 +0200
Subject: [PATCH 24/32] Increase test coverage for aws script.

---
 scripts/aws.py    | 24 +++++---------------
 tests/test_aws.py | 58 +++++++++--------------------------------------
 2 files changed, 17 insertions(+), 65 deletions(-)

diff --git a/scripts/aws.py b/scripts/aws.py
index 0025e44e..0d10701c 100755
--- a/scripts/aws.py
+++ b/scripts/aws.py
@@ -10,16 +10,9 @@ logger = logging.getLogger(__name__)
 
 
 class AWSConnection:
-    def __init__(self, config):
+    def __init__(self, cluster_name):
         self.available = False
-        self.config = config
-
-        if 'cluster_name' in config:
-            self.cluster_name = config.get('cluster_name')
-        elif 'etcd' in config and isinstance(config['etcd'], dict):
-            self.cluster_name = config['etcd'].get('scope', 'unknown')
-        else:
-            self.cluster_name = 'unknown'
+        self.cluster_name = cluster_name if cluster_name is not None else 'unknown'
         try:
             # get the instance id
             r = requests.get('http://169.254.169.254/latest/dynamic/instance-identity/document', timeout=0.1)
@@ -73,12 +66,7 @@ class AWSConnection:
 
 
 if __name__ == '__main__':
-    if len(sys.argv) != 4:
-        print ("Usage: {0} action role name".format(sys.argv[0]))
-        sys.exit(1)
-    action, role, name = sys.argv[1:]
-    if action in ('on_start', 'on_stop', 'on_role_change'):
-        aws = AWSConnection({'cluster_name': name})
-        aws.on_role_change(role)
-        sys.exit(0)
-    sys.exit(2)
+    if len(sys.argv) != 4 and sys.argv[1] in ('on_start', 'on_stop', 'on_role_change'):
+        AWSConnection(cluster_name=sys.argv[3]).on_role_change(sys.argv[2])
+    else:
+        sys.exit("Usage: {0} action role name".format(sys.argv[0]))
diff --git a/tests/test_aws.py b/tests/test_aws.py
index 4331badb..84d495fe 100644
--- a/tests/test_aws.py
+++ b/tests/test_aws.py
@@ -4,7 +4,6 @@ import boto.ec2
 from collections import namedtuple
 from scripts.aws import AWSConnection
 from requests.exceptions import RequestException
-import yaml
 
 
 class MockEc2Connection:
@@ -42,8 +41,8 @@ class TestAWSConnection(unittest.TestCase):
     def set_error(self):
         self.error = True
 
-    def set_ok(self):
-        self.error = False
+    def set_json_error(self):
+        self.json_error = True
 
     def boto_ec2_connect_to_region(self, region):
         return MockEc2Connection(self.error)
@@ -53,7 +52,7 @@ class TestAWSConnection(unittest.TestCase):
             raise RequestException("foo")
         result = namedtuple('Request', 'ok content')
         result.ok = True
-        if url.split('/')[-1] == 'document':
+        if url.split('/')[-1] == 'document' and not self.json_error:
             result = {"instanceId": "012345", "region": "eu-west-1"}
         else:
             result = 'foo'
@@ -61,50 +60,10 @@ class TestAWSConnection(unittest.TestCase):
 
     def setUp(self):
         self.error = False
+        self.json_error = False
         requests.get = self.requests_get
         boto.ec2.connect_to_region = self.boto_ec2_connect_to_region
-        self.config_string = """
-scope: &scope test
-ttl: &ttl 30
-loop_wait: &loop_wait 10
-restapi:
-  listen: 0.0.0.0:8008
-  connect_address: 127.0.0.1:5432
-etcd:
-  scope: *scope
-  ttl: *ttl
-  host: 127.0.0.1:8080
-postgresql:
-  scope: *scope
-  name: postgresql_foo
-  listen: 0.0.0.0:5432
-  connect_address: 127.0.0.1:5432
-  data_dir: /home/postgres/pgdata/data
-  replication:
-    username: standby
-    password: standby
-    network: 0.0.0.0/0
-  superuser:
-    password: zalando
-  admin:
-    username: admin
-    password: admin
-  callbacks:
-    on_start: patroni/scripts/aws.py
-    on_stop: patroni/scripts/aws.py
-    on_restart: patroni/scripts/aws.py
-    on_role_change: patroni/scripts/aws.py
-  parameters:
-    archive_mode: "on"
-    wal_level: hot_standby
-    max_wal_senders: 5
-    wal_keep_segments: 8
-    archive_timeout: 1800s
-    max_replication_slots: 5
-    hot_standby: "on"
-    ssl: "on"
-"""
-        self.conn = AWSConnection(yaml.load(self.config_string))
+        self.conn = AWSConnection('test')
 
     def test_aws_available(self):
         self.assertTrue(self.conn.aws_available())
@@ -116,11 +75,16 @@ postgresql:
 
     def test_non_aws(self):
         self.set_error()
-        conn = AWSConnection(yaml.load(self.config_string))
+        conn = AWSConnection('test')
         self.assertFalse(conn.aws_available())
         self.assertFalse(conn._tag_ebs('master'))
         self.assertFalse(conn._tag_ec2('master'))
 
+    def test_aws_bizare_response(self):
+        self.set_json_error()
+        conn = AWSConnection('test')
+        self.assertFalse(conn.aws_available())
+
     def test_aws_tag_ebs_error(self):
         self.set_error()
         self.assertFalse(self.conn._tag_ebs("master"))

From 7d1bc32ad93d70b1d4814fc0b5bc2badb2aa1130 Mon Sep 17 00:00:00 2001
From: Oleksii Kliukin <oleksii.kliukin@zalando.de>
Date: Mon, 27 Jul 2015 13:06:44 +0200
Subject: [PATCH 25/32] potentially run the doctests for the scripts.

---
 setup.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/setup.py b/setup.py
index db327c37..62d05a69 100644
--- a/setup.py
+++ b/setup.py
@@ -82,7 +82,7 @@ class PyTest(TestCommand):
             params['plugins'] = ['cov']
         if self.junitxml:
             params['args'] += self.junitxml
-        params['args'] += ['--doctest-modules', HELPERS, '-s']
+        params['args'] += ['--doctest-modules', HELPERS, '--doctest-modules', SCRIPTS, '-s']
         errno = pytest.main(**params)
         sys.exit(errno)
 

From 27bda5b4e46237f800bddb373a29b3366fb3eeeb Mon Sep 17 00:00:00 2001
From: Oleksii Kliukin <oleksii.kliukin@zalando.de>
Date: Mon, 27 Jul 2015 13:12:08 +0200
Subject: [PATCH 26/32] document the new callbacks option.

---
 README.md | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/README.md b/README.md
index 3ecbcebb..be12dc71 100644
--- a/README.md
+++ b/README.md
@@ -68,6 +68,12 @@ For an example file, see `postgres0.yml`.  Below is an explanation of settings:
     * *username*: replication username, user will be created during initialization
     * *password*: replication password, user will be created during initialization
     * *network*: network setting for replication in pg_hba.conf
+  * *callbacks* callback scripts to run on certain actions. Patroni will pass current action, role and cluster name. See scripts/aws.py as an example on how to write them.
+    * *on_start*: a script to run when the cluster starts
+    * *on_stop*: a script to run when the cluster stops
+    * *on_restart*: a script to run when the cluster restarts
+    * *on_reload*: a script to run when configuration reload is triggered
+    * *on_role_change*: a script to run when the cluster is being promoted or demoted
   * *recovery_conf*: configuration settings written to recovery.conf when configuring follower
   * *parameters*: list of configuration settings for Postgres
 

From 5c5bae69eaa8b03f893375038424b4db32fac397 Mon Sep 17 00:00:00 2001
From: Oleksii Kliukin <oleksii.kliukin@zalando.de>
Date: Mon, 27 Jul 2015 16:11:58 +0200
Subject: [PATCH 27/32] make sure we can import from scripts.

---
 scripts/__init__.py | 0
 1 file changed, 0 insertions(+), 0 deletions(-)
 create mode 100644 scripts/__init__.py

diff --git a/scripts/__init__.py b/scripts/__init__.py
new file mode 100644
index 00000000..e69de29b

From 71a1200f64cbe60fe2123e19ea1a9035f97615e0 Mon Sep 17 00:00:00 2001
From: Oleksii Kliukin <oleksii.kliukin@zalando.de>
Date: Wed, 29 Jul 2015 12:09:17 +0200
Subject: [PATCH 28/32] Fix a mistake in checking for the number of parameters.

---
 scripts/aws.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/scripts/aws.py b/scripts/aws.py
index b5e87fd7..b172a8c3 100755
--- a/scripts/aws.py
+++ b/scripts/aws.py
@@ -66,7 +66,7 @@ class AWSConnection:
 
 
 if __name__ == '__main__':
-    if len(sys.argv) != 4 and sys.argv[1] in ('on_start', 'on_stop', 'on_role_change'):
+    if len(sys.argv) == 4 and sys.argv[1] in ('on_start', 'on_stop', 'on_role_change'):
         AWSConnection(cluster_name=sys.argv[3]).on_role_change(sys.argv[2])
     else:
         sys.exit("Usage: {0} action role name".format(sys.argv[0]))

From a0fdd6398f0afcf5d29abd89101e9ee3da949b87 Mon Sep 17 00:00:00 2001
From: Oleksii Kliukin <oleksii.kliukin@zalando.de>
Date: Thu, 30 Jul 2015 10:17:17 +0200
Subject: [PATCH 29/32] Cosmetic changes: avoid magic strings in action names,
 name the parameter passed to the callback 'scope' for consistency.

---
 helpers/postgresql.py | 36 +++++++++++++++++++++---------------
 1 file changed, 21 insertions(+), 15 deletions(-)

diff --git a/helpers/postgresql.py b/helpers/postgresql.py
index 325d7080..84a94346 100644
--- a/helpers/postgresql.py
+++ b/helpers/postgresql.py
@@ -16,6 +16,12 @@ else:
 
 logger = logging.getLogger(__name__)
 
+ACTION_ON_START = "on_start"
+ACTION_ON_STOP = "on_stop"
+ACTION_ON_RESTART = "on_restart"
+ACTION_ON_RELOAD = "on_reload"
+ACTION_ON_ROLE_CHANGE = "on_role_change"
+
 
 def parseurl(url):
     r = urlparse(url)
@@ -253,12 +259,12 @@ class Postgresql:
             except psycopg2.OperationalError as e:
                 logger.warning("unable to perform {0} action, cannot obtain the cluster role: {1}".format(cb_name, e))
                 return False
-        name = self.scope
+        scope = self.scope
         try:
             role = "master" if is_leader else "replica"
-            subprocess.Popen(shlex.split(os.path.abspath(cmd))+[cb_name, role, name])
+            subprocess.Popen(shlex.split(os.path.abspath(cmd))+[cb_name, role, scope])
         except Exception as e:
-            logger.warning("callback {0} {1} {2} {3} failed: {4}".format(os.path.abspath(cmd), cb_name, role, name, e))
+            logger.warning("callback {0} {1} {2} {3} failed: {4}".format(os.path.abspath(cmd), cb_name, role, scope, e))
             return False
         return True
 
@@ -275,8 +281,8 @@ class Postgresql:
         ret = subprocess.call(self._pg_ctl + ['start', '-o', self.server_options()]) == 0
         ret and self.load_replication_slots()
         self.save_configuration_files()
-        if ret and 'on_start' in self.callback:
-            self.call_nowait('on_start')
+        if ret and ACTION_ON_START in self.callback:
+            self.call_nowait(ACTION_ON_START)
         return ret
 
     def stop(self):
@@ -286,14 +292,14 @@ class Postgresql:
             is_leader = None
             pass
         ret = subprocess.call(self._pg_ctl + ['stop', '-m', 'fast'])
-        if ret == 0 and 'on_stop' in self.callback:
-            self.call_nowait('on_stop', is_leader=is_leader)
+        if ret == 0 and ACTION_ON_STOP in self.callback:
+            self.call_nowait(ACTION_ON_STOP, is_leader=is_leader)
         return ret == 0
 
     def reload(self):
         ret = subprocess.call(self._pg_ctl + ['reload'])
-        if ret == 0 and 'on_reload' in self.callback:
-            self.call_nowait('on_reload')
+        if ret == 0 and ACTION_ON_RELOAD in self.callback:
+            self.call_nowait(ACTION_ON_RELOAD)
         return ret == 0
 
     def restart(self):
@@ -303,8 +309,8 @@ class Postgresql:
             is_leader = None
             pass
         ret = subprocess.call(self._pg_ctl + ['restart', '-m', 'fast'])
-        if ret == 0 and 'on_restart' in self.callback:
-            self.call_nowait('on_restart', is_leader=is_leader)
+        if ret == 0 and ACTION_ON_RESTART in self.callback:
+            self.call_nowait(ACTION_ON_RESTART, is_leader=is_leader)
         return ret == 0
 
     def server_options(self):
@@ -392,8 +398,8 @@ primary_conninfo = '{}'
         if not self.check_recovery_conf(leader):
             self.write_recovery_conf(leader)
             self.restart()
-            if 'on_role_change' in self.callback:
-                self.call_nowait('on_role_change')
+            if ACTION_ON_ROLE_CHANGE in self.callback:
+                self.call_nowait(ACTION_ON_ROLE_CHANGE)
 
     def save_configuration_files(self):
         """
@@ -413,8 +419,8 @@ primary_conninfo = '{}'
 
     def promote(self):
         self.is_promoted = subprocess.call(self._pg_ctl + ['promote']) == 0
-        if self.is_promoted and 'on_role_change' in self.callback:
-            self.call_nowait('on_role_change')
+        if self.is_promoted and ACTION_ON_ROLE_CHANGE in self.callback:
+            self.call_nowait(ACTION_ON_ROLE_CHANGE)
         return self.is_promoted
 
     def demote(self, leader):

From 7c4efb33e74a2b715b964011566ae2e7a7d0c6ad Mon Sep 17 00:00:00 2001
From: Feike Steenbergen <feikesteenbergen@gmail.com>
Date: Wed, 5 Aug 2015 09:56:32 +0200
Subject: [PATCH 30/32] Updated Dockerfile

Due to the renaming of Governor to Patroni some old references needed to be updated.
Also some python packages need to be added.

Added entrypoint.sh as a script, to ensure Patroni will have PID = 1 when the container is run.
---
 Dockerfile    | 19 ++++++++++---------
 entrypoint.sh |  3 +++
 2 files changed, 13 insertions(+), 9 deletions(-)
 create mode 100755 entrypoint.sh

diff --git a/Dockerfile b/Dockerfile
index d44a9af3..d8e27c2d 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -1,4 +1,4 @@
-## This Dockerfile is meant to aid in the building and debugging governor whilst developing on your local machine
+## This Dockerfile is meant to aid in the building and debugging patroni whilst developing on your local machine
 ## It has all the necessary components to play/debug with a single node appliance, running etcd
 FROM ubuntu:14.04
 MAINTAINER Feike Steenbergen <feike.steenbergen@zalando.de>
@@ -13,22 +13,23 @@ RUN apt-get update -y
 RUN apt-get upgrade -y
 
 ENV PGVERSION 9.4
-RUN apt-get install python python-psycopg2 python-yaml python-requests python-boto postgresql-${PGVERSION} python-dnspython -y
+RUN apt-get install python python-psycopg2 python-yaml python-requests python-boto postgresql-${PGVERSION} python-dnspython python-pip -y
+RUN pip install zake
 
 ENV PATH /usr/lib/postgresql/${PGVERSION}/bin:$PATH
 
-RUN mkdir -p /governor/helpers
-ADD governor.py /governor/governor.py
-ADD helpers /governor/helpers
-ADD postgres0.yml /governor/
+RUN mkdir -p /patroni/helpers
+ADD patroni.py /patroni/patroni.py
+ADD helpers /patroni/helpers
+ADD postgres0.yml /patroni/
 
 ENV ETCDVERSION 2.0.12
 RUN curl -L https://github.com/coreos/etcd/releases/download/v${ETCDVERSION}/etcd-v${ETCDVERSION}-linux-amd64.tar.gz | tar xz -C /bin --strip=1 --wildcards --no-anchored etcd etcdctl
 
 ## Setting up a simple script that will serve as an entrypoint
 RUN mkdir /data/ && touch /var/log/etcd.log /var/log/etcd.err && chown postgres:postgres /var/log/etcd.*
-RUN chown postgres:postgres -R /governor/ /data/
-RUN /bin/echo -e "etcd --data-dir /tmp/etcd.data > /var/log/etcd.log 2> /var/log/etcd.err &\n/governor/governor.py /governor/postgres0.yml \"\$@\"" >> /entrypoint.sh && chmod +x /entrypoint.sh
+RUN chown postgres:postgres -R /patroni/ /data/
+ADD entrypoint.sh /entrypoint.sh
 
-ENTRYPOINT /entrypoint.sh
+ENTRYPOINT ["/bin/bash", "/entrypoint.sh"]
 USER postgres
diff --git a/entrypoint.sh b/entrypoint.sh
new file mode 100755
index 00000000..ab76a0dd
--- /dev/null
+++ b/entrypoint.sh
@@ -0,0 +1,3 @@
+#!/bin/bash
+etcd --data-dir /tmp/etcd.data > /var/log/etcd.log 2> /var/log/etcd.err &
+exec /patroni/patroni.py /patroni/postgres0.yml "$@"

From c30d8dbd1ae278795e36d680999e8f8d0b1f4e4a Mon Sep 17 00:00:00 2001
From: Feike Steenbergen <feikesteenbergen@gmail.com>
Date: Wed, 5 Aug 2015 14:12:45 +0200
Subject: [PATCH 31/32] Development: Update Dockerfile and create possibility
 to run local cluster

To help in developing features, the Dockerfile and its entrypoint have been extended.
The README.md explains stuff in detail, in short:

- you can now run a Patroni cluster with a single command
---
 Dockerfile                    |  14 ++--
 docker/README.md              |  46 +++++++++++++
 docker/dev_patroni_cluster.sh |  90 +++++++++++++++++++++++++
 docker/entrypoint.sh          | 122 ++++++++++++++++++++++++++++++++++
 entrypoint.sh                 |   3 -
 5 files changed, 265 insertions(+), 10 deletions(-)
 create mode 100644 docker/README.md
 create mode 100755 docker/dev_patroni_cluster.sh
 create mode 100755 docker/entrypoint.sh
 delete mode 100755 entrypoint.sh

diff --git a/Dockerfile b/Dockerfile
index d8e27c2d..147cad38 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -13,23 +13,23 @@ RUN apt-get update -y
 RUN apt-get upgrade -y
 
 ENV PGVERSION 9.4
-RUN apt-get install python python-psycopg2 python-yaml python-requests python-boto postgresql-${PGVERSION} python-dnspython python-pip -y
-RUN pip install zake
+RUN apt-get install python python-psycopg2 python-yaml python-requests python-boto postgresql-${PGVERSION} python-dnspython python-kazoo -y
 
 ENV PATH /usr/lib/postgresql/${PGVERSION}/bin:$PATH
 
 RUN mkdir -p /patroni/helpers
 ADD patroni.py /patroni/patroni.py
 ADD helpers /patroni/helpers
-ADD postgres0.yml /patroni/
 
-ENV ETCDVERSION 2.0.12
+ENV ETCDVERSION 2.0.13
 RUN curl -L https://github.com/coreos/etcd/releases/download/v${ETCDVERSION}/etcd-v${ETCDVERSION}-linux-amd64.tar.gz | tar xz -C /bin --strip=1 --wildcards --no-anchored etcd etcdctl
 
 ## Setting up a simple script that will serve as an entrypoint
-RUN mkdir /data/ && touch /var/log/etcd.log /var/log/etcd.err && chown postgres:postgres /var/log/etcd.*
-RUN chown postgres:postgres -R /patroni/ /data/
-ADD entrypoint.sh /entrypoint.sh
+RUN mkdir /data/ && touch /var/log/etcd.log /var/log/etcd.err /pgpass /patroni/postgres.yml
+RUN chown postgres:postgres -R /patroni/ /data/ /pgpass /var/log/etcd.* /patroni/postgres.yml
+ADD docker/entrypoint.sh /entrypoint.sh
+
+EXPOSE 4001 5432 2380
 
 ENTRYPOINT ["/bin/bash", "/entrypoint.sh"]
 USER postgres
diff --git a/docker/README.md b/docker/README.md
new file mode 100644
index 00000000..e6245adc
--- /dev/null
+++ b/docker/README.md
@@ -0,0 +1,46 @@
+# Patroni Dockerfile
+You can run Patroni in a docker container using this Dockerfile, or by using the Docker image at
+    https://os-registry.stups.zalan.do/acid/patroni-1.0
+
+This Dockerfile is meant in aiding development of Patroni and quick testing of features. It is not a production-worthy
+Dockerfile
+
+# Examples
+
+## Standalone Patroni
+
+    docker run -d os-registry.stups.zalan.do/acid/patroni:1.0
+
+## Multiple Patroni's communicating with a standalone etcd inside Docker
+
+Basically what you would do would be: 
+
+* Run 1 container which provides etcd
+
+        docker run -d <IMAGE> --etcd-only
+
+* Run n containers running Patroni, passing the `--etcd` option to the `docker run` command
+
+        docker run -d <IMAGE> --etcd=<IP FROM etcd CONTAINER:PORT>
+
+To automate this you can run the following script:
+
+    dev_patroni_cluster.sh [OPTIONS]
+
+    Options:
+
+        --image     IMAGE    The Docker image to use for the cluster
+        --members   INT      The number of members for the cluster
+        --name      NAME     The name of the new cluster
+
+Example session:
+
+    $ ./dev_patroni_cluster.sh --image os-registry.stups.zalan.do/acid/patroni:1.0 --members=2 --name=bravo
+    The etcd container is 6be871a11cb373406ca5ea1c6b39e140fdde9fb1d6177212d6ad0c0d1bd9b563, ip=172.17.1.24
+    Started Patroni container 67e611f2eca7c40f9e6e0e24a4a8f2cba7e3e56d22a420e15ab9240a37a9d7a4, ip=172.17.1.25
+    Started Patroni container 47dd12ae635ab83b039f5889e250048b606ed5e48e3650b69e365e7e1d4acbcf, ip=172.17.1.26
+    $ docker ps
+    CONTAINER ID        IMAGE                                         COMMAND                CREATED             STATUS              PORTS                          NAMES
+    47dd12ae635a        os-registry.stups.zalan.do/acid/patroni:1.0   "/bin/bash /entrypoi   10 seconds ago      Up 8 seconds        4001/tcp, 5432/tcp, 2380/tcp   bravo_OR64g8bx
+    67e611f2eca7        os-registry.stups.zalan.do/acid/patroni:1.0   "/bin/bash /entrypoi   11 seconds ago      Up 10 seconds       2380/tcp, 4001/tcp, 5432/tcp   bravo_si9no8iz
+    6be871a11cb3        os-registry.stups.zalan.do/acid/patroni:1.0   "/bin/bash /entrypoi   12 seconds ago      Up 10 seconds       4001/tcp, 5432/tcp, 2380/tcp   bravo_etcd
diff --git a/docker/dev_patroni_cluster.sh b/docker/dev_patroni_cluster.sh
new file mode 100755
index 00000000..00a8264a
--- /dev/null
+++ b/docker/dev_patroni_cluster.sh
@@ -0,0 +1,90 @@
+#!/bin/bash
+
+DOCKER_IMAGE="os-registry.stups.zalan.do/acid/patroni:1.0"
+MEMBERS=3
+
+
+function usage()
+{
+    cat <<__EOF__
+Usage: $0 
+
+Options:
+
+    --image     IMAGE    The Docker image to use for the cluster
+    --members   INT      The number of members for the cluster
+    --name      NAME     The name of the new cluster
+
+Examples:
+
+    $0 --image ${DOCKER_IMAGE}
+    $0
+    $0 --image ${DOCKER_IMAGE} --members=2
+__EOF__
+}
+
+
+optspec=":-:"
+while getopts "$optspec" optchar; do
+    case "${optchar}" in
+        -)
+            case "${OPTARG}" in
+                help)
+                    usage
+                    exit 0
+                    ;;
+                name)
+                    PATRONI_SCOPE="${!OPTIND}"; OPTIND=$(( $OPTIND + 1 ))
+                    ;;
+                name=*)
+                    PATRONI_SCOPE="${OPTARG#*=}"
+                    ;;
+                image)
+                    DOCKER_IMAGE="${!OPTIND}"; OPTIND=$(( $OPTIND + 1 ))
+                    ;;
+                image=*)
+                    DOCKER_IMAGE="${OPTARG#*=}"
+                    ;;
+                members)
+                    MEMBERS="${!OPTIND}"; OPTIND=$(( $OPTIND + 1 ))
+                    ;;
+                members=*)
+                    MEMBERS="${OPTARG#*=}"
+                    ;;
+                *)
+                    if [ "$OPTERR" = 1 ] && [ "${optspec:0:1}" != ":" ]; then
+                        echo "Unknown option --${OPTARG}" >&2
+                    fi
+                    ;;
+            esac;;
+        *)
+            if [ "$OPTERR" != 1 ] || [ "${optspec:0:1}" = ":" ]; then
+                echo "Non-option argument: '-${OPTARG}'" >&2
+                usage
+                exit 1
+            fi
+            ;;
+    esac
+done
+
+function random_name()
+{
+    cat /dev/urandom | env LC_CTYPE=C tr -dc 'a-zA-Z0-9' | head -c 8
+}
+
+if [ -z ${PATRONI_SCOPE} ]
+then
+    PATRONI_SCOPE=$(random_name)
+fi
+
+etcd_container=$(docker run -d --name="${PATRONI_SCOPE}_etcd" "${DOCKER_IMAGE}" --etcd-only)
+etcd_container_ip=$(docker inspect --format '{{ .NetworkSettings.IPAddress }}' ${etcd_container})
+echo "The etcd container is ${etcd_container}, ip=${etcd_container_ip}"
+
+for i in $(seq 1 "${MEMBERS}")
+do
+    container_name=$(random_name)
+    patroni_container=$(docker run -d --name="${PATRONI_SCOPE}_${container_name}" "${DOCKER_IMAGE}" --etcd="${etcd_container_ip}:4001" --name="${PATRONI_SCOPE}")
+    patroni_container_ip=$(docker inspect --format '{{ .NetworkSettings.IPAddress }}' ${patroni_container})
+    echo "Started Patroni container ${patroni_container}, ip=${patroni_container_ip}"
+done
diff --git a/docker/entrypoint.sh b/docker/entrypoint.sh
new file mode 100755
index 00000000..697bab66
--- /dev/null
+++ b/docker/entrypoint.sh
@@ -0,0 +1,122 @@
+#!/bin/bash
+
+function usage()
+{
+    cat <<__EOF__
+Usage: $0 
+
+Options:
+
+    --etcd      ETCD    Provide an external etcd to connect to
+    --name      NAME    Give the cluster a specific name 
+    --etcd-only         Do not run Patroni, run a standalone etcd
+
+Examples:
+
+    $0 --etcd=127.17.0.84:4001
+    $0 --etcd-only
+    $0 
+    $0 --name=true_scotsman
+__EOF__
+}
+
+DOCKER_IP=$(hostname --ip-address)
+PATRONI_SCOPE=batman
+
+optspec=":vh-:"
+while getopts "$optspec" optchar; do
+    case "${optchar}" in
+        -)
+            case "${OPTARG}" in
+                etcd-only)
+                    exec etcd --data-dir /tmp/etcd.data \
+                        -advertise-client-urls=http://${DOCKER_IP}:4001 \
+                        -listen-client-urls=http://0.0.0.0:4001 \
+                        -listen-peer-urls=http://0.0.0.0:2380 
+                    exit 0
+                    ;;
+                name)
+                    PATRONI_SCOPE="${!OPTIND}"; OPTIND=$(( $OPTIND + 1 ))
+                    ;;
+                name=*)
+                    PATRONI_SCOPE=${OPTARG#*=}
+                    ;;
+                etcd)
+                    ETCD_CLUSTER="${!OPTIND}"; OPTIND=$(( $OPTIND + 1 ))
+                    ;;
+                etcd=*)
+                    ETCD_CLUSTER=${OPTARG#*=}
+                    ;;
+                help)
+                    usage
+                    exit 0
+                    ;;
+                *)
+                    if [ "$OPTERR" = 1 ] && [ "${optspec:0:1}" != ":" ]; then
+                        echo "Unknown option --${OPTARG}" >&2
+                    fi
+                    ;;
+            esac;;
+        *)
+            if [ "$OPTERR" != 1 ] || [ "${optspec:0:1}" = ":" ]; then
+                echo "Non-option argument: '-${OPTARG}'" >&2
+                usage
+                exit 1
+            fi
+            ;;
+    esac
+done
+
+if [ -z ${ETCD_CLUSTER} ]
+then
+    etcd --data-dir /tmp/etcd.data > /var/log/etcd.log 2> /var/log/etcd.err &
+    ETCD_CLUSTER="127.0.0.1:4001"
+fi
+
+cat > /patroni/postgres.yml <<__EOF__
+
+ttl: &ttl 30
+loop_wait: &loop_wait 10
+scope: &scope ${PATRONI_SCOPE}
+restapi:
+  listen: 127.0.0.1:8008
+  connect_address: 127.0.0.1:8008
+etcd:
+  scope: *scope
+  ttl: *ttl
+  host: ${ETCD_CLUSTER}
+postgresql:
+  name: postgresql_${DOCKER_IP//./_} ## Replication slots do not allow dots in their name
+  scope: *scope
+  listen: 0.0.0.0:5432
+  connect_address: ${DOCKER_IP}:5432
+  data_dir: data/postgresql0
+  maximum_lag_on_failover: 1048576 # 1 megabyte in bytes
+  pg_hba:
+  - host all all 0.0.0.0/0 md5
+  - hostssl all all 0.0.0.0/0 md5
+  - host replication replicator ${DOCKER_IP}/16    md5
+  replication:
+    username: replicator
+    password: rep-pass
+    network:  127.0.0.1/32
+  superuser:
+    password: zalando
+  admin:
+    username: admin
+    password: admin
+  parameters:
+    archive_mode: "on"
+    wal_level: hot_standby
+    archive_command: mkdir -p ../wal_archive && cp %p ../wal_archive/%f
+    max_wal_senders: 20 
+    listen_addresses: 0.0.0.0
+    wal_keep_segments: 8
+    archive_timeout: 1800s
+    max_replication_slots: 20
+    hot_standby: "on"
+__EOF__
+
+cat /patroni/postgres.yml
+
+exec /patroni/patroni.py /patroni/postgres.yml
diff --git a/entrypoint.sh b/entrypoint.sh
deleted file mode 100755
index ab76a0dd..00000000
--- a/entrypoint.sh
+++ /dev/null
@@ -1,3 +0,0 @@
-#!/bin/bash
-etcd --data-dir /tmp/etcd.data > /var/log/etcd.log 2> /var/log/etcd.err &
-exec /patroni/patroni.py /patroni/postgres0.yml "$@"

From cdb0e43ed771fcf3fceffc587735aba99523e0a6 Mon Sep 17 00:00:00 2001
From: Feike Steenbergen <feikesteenbergen@gmail.com>
Date: Wed, 5 Aug 2015 16:53:05 +0200
Subject: [PATCH 32/32] Dockerfile: Enable (undocumented) cheat mode to
 troubleshoot

---
 docker/entrypoint.sh | 18 ++++++++++++++++--
 1 file changed, 16 insertions(+), 2 deletions(-)

diff --git a/docker/entrypoint.sh b/docker/entrypoint.sh
index 697bab66..f46e7ea5 100755
--- a/docker/entrypoint.sh
+++ b/docker/entrypoint.sh
@@ -35,6 +35,9 @@ while getopts "$optspec" optchar; do
                         -listen-peer-urls=http://0.0.0.0:2380 
                     exit 0
                     ;;
+                cheat)
+                    CHEAT=1
+                    ;;
                 name)
                     PATRONI_SCOPE="${!OPTIND}"; OPTIND=$(( $OPTIND + 1 ))
                     ;;
@@ -69,7 +72,10 @@ done
 
 if [ -z ${ETCD_CLUSTER} ]
 then
-    etcd --data-dir /tmp/etcd.data > /var/log/etcd.log 2> /var/log/etcd.err &
+    etcd --data-dir /tmp/etcd.data \
+        -advertise-client-urls=http://${DOCKER_IP}:4001 \
+        -listen-client-urls=http://0.0.0.0:4001 \
+        -listen-peer-urls=http://0.0.0.0:2380 > /var/log/etcd.log 2> /var/log/etcd.err &
     ETCD_CLUSTER="127.0.0.1:4001"
 fi
 
@@ -119,4 +125,12 @@ __EOF__
 
 cat /patroni/postgres.yml
 
-exec /patroni/patroni.py /patroni/postgres.yml
+if [ ! -z $CHEAT ]
+then
+    while :
+    do
+        sleep 60
+    done
+else
+    exec /patroni/patroni.py /patroni/postgres.yml
+fi